the agony and the ecstasy ☠️

I ain’t happy, I’m feeling glad I got sunshine in a bag I’m useless, but not for long The future is coming on I ain’t happy, I’m feeling glad I got sunshine in a bag I’m useless, but not for long The future is coming on (That’s right) It’s coming on, it’s coming on It’s coming on, it’s coming on

She paints her nails and she don’t know He’s got her best friend on the phone She’ll wash her hair His dirty clothes are all he gives to her And he’s got posters on the wall Of all the girls he wished she was And he means everything to her

Her boyfriend, he don’t know Anything about her He’s too stoned, Nintendo I wish that I could make her see She’s just the flavor of the weak

To teenage me, taking punk rock and turning it into poppy music about boy/girl romance and life in the suburbs was "defanging" of the genre. I used to rant that it was inevitable - since the invention of recorded music, whenever a genre challenged the status quo, industry gatekeepers would eventually start selling back a diluted version of it, and the establishment would profit off a movement meant to criticize it. I would've argued that Elvis, the Beatles, disco, hair metal, grunge, and then emo were all commercialized forms of music that had originally challenged mainstream American society, reframed as radio-friendly love songs, sanitized to the point of self-parody. Teenage me was kind of an edgelord buzzkill.

The condensed version of this argument was singing "too stoned, Nintendo" in a whiny falsetto.

All I know is that to me You look like you’re havin’ fun Open up your lovin’ arms Watch out, here I come

You spin me right ‘round, baby, right ‘round Like a record, baby, right 'round, 'round, 'round You spin me right 'round, baby, right 'round Like a record, baby, right 'round, 'round, 'round

(rip flash)

original url http://www.geocities.com/SunsetStrip/Pit/3210/

By the time you hear the siren It’s already too late One goes to the morgue and the other to jail One guy’s wasted and the other’s a waste It goes down the same as the thousand before No one’s getting smarter No one’s learning the score Your never-ending spree of death and violence and hate Is gonna tie your own rope, tie your own rope, tie your own…

It seems like a small thing, but I'd always considered myself as a person that didn't "like" vegetables. The larger conversation was pointing out that there were lots of vegetables I willingly ate and even enjoyed already, so advice like "eat more vegetables" didn't have to mean steamed broccoli and kale smoothies - things I can't stand. Onions and jalapenos are still vegetables, even rolled up in a burrito. Baby spinach is still a rich leafy green even if it's on top of a bacon cheeseburger.

1. The company goes under because it can’t keep tricking its investors forever (extremely common with most free or cheap startups)

2. The company is forced to turn into an advertising business to survive (early days Google, most social media sites, etc.)

3. The company drives out all other competition and then raises its prices once it’s the only viable option left and its customers are trapped with it (Amazon, uber, etc.)

Now here’s the thing I can’t figure out, which is what all of these AI companies are gonna turn out to be. Because investors are seemingly holding onto this AI thing very tightly, it’s unclear how they could be an advertising company unless they’re selling their products to advertisers who can already use their products for free, and they’re kind of filling an unnecessary niche. I mean, a lot of people clearly find value in AI chatbots even if I don’t like them, but is that enough value for them to be forced to pay for it? I don’t think so. I think for most people it’s like a novelty they can interact with and like every company has a large language model these days so if you start charging for it they can just move to Google or Microsoft’s AI which are large corporations that don’t need to charge extra for the use of their AI.

This is why a lot of people think we’re in an AI bubble and that the bubble’s gonna pop, taking half of the investment market with it. There’s just no way for these individual companies to become profitable in the long term unless they choose to be bought and become subsidiaries of larger companies, but those larger companies have the resources to just create their own LLMs so why would they? Not to mention all of the copyright issues.

The thing is though is that the bubble hasn’t popped yet. Which is weird. It’s pretty obvious to like a lot of people across a lot of industries including the finance industry that an AI large language model isn’t really profitable in isolation, especially if you get people used to using it for free.

It makes me wonder how long until the bubble bursts. It might deflate soon at the very least because currently they’re building more square footage of data centers than is physically needed for basically anything, including generative AI. A lot of communities are also protesting the building of these large data centers in their regions because they’re loud and they bother residents and scare off local wildlife and don’t even create a lot of jobs.

You really don’t need a lot of people to run a data center so once all of the data centers are built and the construction jobs dry up and they have more square footage than they can use there’s not gonna be a good economic impact from all of this and the noise pollution is gonna piss people off.

a) "Sponsored results" that have little or nothing to do with your search terms, baked right in at the top of the list in almost every app or platform that offers a search feature

b) Elon's highly public and unsubtle manipulation of Grok's responses on various subjects (and the userbase's complete lack of concern about a single person being able to control access to information in that way)

c) The general tactics of telecom providers and tech companies over the last 50 years

My prediction is that the monetization of LLMs will come from people paying to make it lie. Right now we're in the honeymoon phase where the user experience and factual accuracy are priorities. But:

a) People have proven they're willing to uncritically accept advice from LLMs that are wrong 30%-40% of the time, and

b) The data centers will benefit existing LLM providers by creating a high barrier to entry shutting out competition. Data centers are expensive, making the initial investment to build one a roadblock to new startups anyway. Then the more public resistance there is to new ones being built, the more valuable existing data centers will become.

The enshittification will follow a pattern:

a) LLM providers will stop trying to make it any more accurate, users have deemed it "good enough," so further improvement is diminishing returns for investors

b) Anyone with a product to sell or an agenda to push will be able to pay to affect the LLM's responses. Just like Google AdWords, controversial topics and terms related to highly competitive industries will fetch the highest prices. Asking ChatGPT about your symptoms will give you a sponsored answer from a pharmaceutical company selling a drug to treat them. Asking Claude about what car is the most reliable will give you a sponsored answer from Toyota. Asking about election polling will give you an answer sponsored by whichever candidate had the most money to spend on altering the response. The LLMs may or may not be transparent that you're being advertised to. It will likely be swept under the rug as something like "better accuracy with replies on complex topics specially crafted by domain experts."